w a t e r r e s e a r c h 4 2 ( 2 0 0 8 ) 3 6 8 5 – 3 6 9 6

Avai lab le a t www.sc iencedi rec t .com

journa l homepage : www.e lsev ie r . com/ loca te /wat res

Evaluation of water quality in an agricultural watershedas affected by almond pest management practices

Xuyang Zhanga, Xingmei Liua,b, Yuzhou Luoa, Minghua Zhanga,c,*aDepartment of Land, Air and Water Resources, University of California, Davis, 1 Shields Avenue, CA 95616, USAbInstitute of Soil, Water and Environmental Sciences, Zhejiang University, Hangzhou 310029, ChinacCalifornia Department of Pesticide Regulation, 1001 I Street, Sacramento, CA 95814, USA

a r t i c l e i n f o

Article history:

Received 22 February 2008

Received in revised form

16 May 2008

Accepted 20 May 2008

Available online 24 June 2008

Keywords:

Water quality

Pest management practices

Organophosphate

Pyrethroid

SWAT

* Corresponding author. Department of LandTel.: þ1 530 752 4953; fax: þ1 530 752 5262.

E-mail addresses: mzhang@cdpr.ca.gov, m0043-1354/$ – see front matter ª 2008 Elsevidoi:10.1016/j.watres.2008.05.018

a b s t r a c t

In the last decade, the detection of organophosphate (OP) pesticides in the San Joaquin

River watershed has raised concerns about water quality. This study examined the influ-

ences of almond pest management practices (PMPs) on water quality. The Soil and Water

Assessment Tool (SWAT) model was employed to simulate pesticide concentration in

water as affected by different PMPs. California Pesticide Use Reporting (PUR) data were

used to investigate PMP use trends. Stepwise regression analysis was performed to test

the correlation between specific PMP use and pesticide concentrations in surface water

and sediment. Our results showed an increasing use of reduced risk pesticides and pyre-

throids on almonds. SWAT simulation over the period of 1992–2005 showed decreases in

OP concentrations in surface water. High OP and pyrethroid use in dormant sprays was

associated with high pesticide concentrations in water and sediment. Almond pesticide

use was proved to have significant impacts on the pesticide load in the San Joaquin River

watershed. The PMP which combines the use of reduced risk pesticides with no dormant

spray was recommended for almond orchard use. This paper presented a novel method

of studying the environmental impacts of different agricultural PMPs. By combining pesti-

cide use surveys with watershed modeling, we provided a quantitative foundation for the

selection of PMPs to reduce pesticide pollution in surface water.

ª 2008 Elsevier Ltd. All rights reserved.

1. Introduction is from December to February, coincides with the rainy season

California produces 99% of the US almond crop, 58% of which

was located in the San Joaquin Valley in 2006 (CDFA and

USDA, 2006). Major pests of almonds include navel orange

worm (NOW), San Jose scale (SJS), peach twig borer (PTB),

and European red and brown mite. Pest management prac-

tices (PMPs), such as the application of organophosphate

(OP) pesticides and oil mixtures during the dormant season,

are considered effective in controlling these pests (Rice

et al., 1972; UCIPM, 1985). However, the dormant period, which

, Air and Water Resources

hzhang@ucdavis.edu (Mer Ltd. All rights reserved

in California potentially causing off-site movement of OP pes-

ticides to water bodies (Zhang et al., 2005; Bacey et al., 2005;

Guo, 2003; Dubrovsky et al., 1998).

OPs such as diazinon and chlorpyrifos have been routinely

detected in the surface water bodies of the San Joaquin River

(SJR) watershed during the rainy season (Spurlock, 2002;

Domagalski et al., 1997). Studies have indicated that runoff

from orchards is a source of these OPs (Domagalski et al.,

1997). Although aerial drift contributes to off-site movement,

surface runoff is the main pathway by which OP pesticides

, University of California, Davis, 1 Shields Avenue, CA 95616, USA.

. Zhang)..

w a t e r r e s e a r c h 4 2 ( 2 0 0 8 ) 3 6 8 5 – 3 6 9 63686

are transported to the SJR (California Regional Water Quality

Control Board Central Valley Region, 2006). Furthermore, re-

cent findings revealed adverse effects of OPs on many species

dependent on surface water quality for survival, such as the

California red-legged frog, California tiger salamander, Delta

smelt and many birds (Miller, 2006; Ross et al., 1999; Spurlock,

2002; Werner et al., 2002). The frequent detection of OP insec-

ticides and their toxicity to species dependent on good water

quality has resulted in increased regulation of OP use.

As a result of these increased restrictions on OP use,

researchers and regulators have been in search of viable alter-

native pest management practices. The California Department

of Pesticide Regulation (Elliott et al., 2004) and UC Statewide

Integrated Pest Management Program (UC IPM) have proposed

several alternative PMPs to the currently used dormant OP

applications (Zalom et al., 1999), such as: (1) no dormant treat-

ment with in-season sprays as needed (no dormant use); (2)

dormant oil alone; (3) bloomtime sprays for peach twig borer;

(4) reduced risk pesticides (e.g. spinosad) as a dormant spray;

(5) conventional non-OP pesticides (e.g. pyrethroids); and (6)

pheromone mating disruption. In addition, to promote the

use of reduced risk pesticides, the US Environmental Protec-

tion Agency (USEPA) Office of Pesticide Programs proposed

a list of more than 170 reduced risk or OP alternative pesticide

products (USEPA, 2006). Despite these efforts, however, little

work has been done to quantitatively evaluate the influence

of the proposed PMPs and EPA-listed products on surface water

quality, especially at watershed scale.

Additionally, there is a lack of literature addressing recent

changes in almond pest management practices and their

influence on surface water quality. Two studies have investi-

gated the historical changes of PMPs. Epstein et al. (2001)

used the Pesticide Use Reporting (PUR) data during 1992–

1997 to investigate the changes in pest management practice

in almond orchards during the dormant season. They found

that the use of OPs decreased both in the acreage treated

and by the percent of growers applying OP products, while

the use of pyrethroid, BT, oil alone, and no dormant treatment

increased. Another study further confirmed the decreasing

trend of dormant OP use by analyzing the PUR data from

1992 to 2000 (Zhang et al., 2005). While these two studies pro-

vided important documentation on the changes of almond

PMPs, they did not, however, cover the more recent changes

in the years since 2000. In addition, none of the past research

has studied the influence of almond PMPs on the environ-

ment, especially on surface water quality. Therefore, the two

objectives of this paper were to (1) identify the use trends of

traditional and alternative PMPs from 1992 to 2005; and (2)

investigate the environmental impacts of the PMPs on surface

water quality.

2. Materials and methods

2.1. Study area

The SJR watershed is an important agricultural production

area located in the Central Valley of California, responsible

for the drainage of 19,000 km2 of primarily agricultural lands

(Fig. 1). There are about 1480 km2 of almond orchards within

the SJR watershed, on which an annual average of 55,728 kg

of OPs was used during 1992–2005 (CDPR PUR database,

2006). About half of the OPs were used in the dormant season.

Many of the pesticides were sprayed on orchards close to

impaired water bodies, which are listed on the Clean Water

303 d list due to the detections of two OP active ingredients,

diazinon and chlorpyrifos. The considerable amount of pesti-

cide use combined with the orchard proximity to the impaired

water bodies (Fig. 1) points to almond orchards as being

important potential sources of agricultural pesticide runoff.

Since the SJR watershed and surface water bodies support

municipal, industrial and agricultural water use in addition

to providing habitat for fish and wildlife species, it is crucial

to reduce the risk to this watershed of OP runoff from almond

orchards.

2.2. Data sources

Pesticide use information from 1992 to 2005 was obtained

from the PUR database maintained by the California Depart-

ment of Pesticide Regulation (CDPR, 2006). The database re-

cords pesticide use information on every application of

a pesticide in production agriculture and also applications

from some non-agricultural entities. The database includes

information on the amount of pesticide product used, the

amount of active ingredient used, the application date, the

planted area and the treated area. The following measures

were used to assess the use trends of PMP. (1) Total reported

amount of applied active ingredient (AI) (kg of AI); (2) total

area treated from all applications even if the same field was

treated more than once (ha treated); and (3) number of almond

growers reporting use of a particular pesticide.

This study acquired other data for watershed modeling

from various databases maintained by federal and state

agencies. Weather data from 1990 to 2005 were obtained

from the California Irrigation Management Information Sys-

tem (CIMIS, 2005). Soil data were obtained from the State

Soil Survey Geographic database (SSURGO) maintained by

the Natural Resources Conservation Service (NRCS, 2006).

Land use data was obtained from EPA’s Land Use and Land

Cover (LULC) spatial data set (USEPA, 2007). Measured data

of stream flow, sediment load and pesticide concentration

were obtained from the National Water Information System

(NWIS, 2007) maintained by USGS for the monitoring site in

the San Joaquin River near Vernalis, California (USGS site ID

11303500).

2.3. PUR data quality

Although the PUR database is the best database available to

reflect pesticide use in California, CDPR has developed error

checking procedures which identify outliers and errors in

variables including rates of use, grower identification and

site location identification (Wilhoit et al., 2001). An error was

identified if the record was considered to be a duplicate or if

the unit for treated areas was anything other than square

feet or acres. Once identified as an error, the record was

deleted. An outlier was identified if the use rate was greater

than (1) 1.12 kg ha�1 (200 lbs per acre) treated; (2) 50 times

the median kg per ha treated for all uses of that product on

Fig. 1 – San Joaquin River Watershed.

w a t e r r e s e a r c h 4 2 ( 2 0 0 8 ) 3 6 8 5 – 3 6 9 6 3687

almonds; and (3) a threshold value determined by a neural

network (Wilhoit et al., 2001). Identified outliers were replaced

with the median use rate (kg of AI per ha treated) of the same

product on almond in the same year. For example, if a record

in 1998 was identified as an outlier by the criteria described

above, the use rate of this record denoted as Rate (old) will

be replaced with the median rate of all the records on the

same product in 1998 denoted as Rate (new). Therefore, the

new value of kilograms of AI denoted as kg of AI (new) was cal-

culated using the following equation

kg of AIðnewÞ ¼ kg of AIðoldÞkg of productðoldÞ � RateðnewÞ � treated ha

(1)

The data cleaning processes identified 2386 records as

errors and 2694 as outliers out of total 625,875 records on

almonds from 1992 to 2005. Therefore, errors and outliers

account for 0.81% of the total records queried, indicating

that the raw data in the database are of good quality.

2.4. PMP scenarios

In addition to traditionally used OP and oil combinations, pes-

ticides used by almond growers in the past 14 years include

pyrethroids, reduced risk pesticides (as identified by the

Crop Protection Handbook, Meister and Sine, 2005), Bacillus

thuringiensis (BT) pesticides, carbamates (Carb), pheromones,

and other pesticides that do not belong to any of the previous

groups (see Table S1 in the Supplemental Information). OPs,

listed in descending order by the total amount of use during

1992–2005, included chlorpyrifos, diazinon, azinphos-methyl,

phosmet, methidathion, naled, parathion, malathion, fenami-

phos, methyl parathion ethoprop, dimethoate, acephate,

phosalone, dicrotophos, and disulfoton. Pyrethrins and

pyrethroids (PYs), in descending order of use, included per-

methrin, esfenvalerate, lambda-cyhalothrin, pyrethrins,

cypermethrin, cyfluthrin, and tau-fluvalinate. Reduced risk

pesticides included tebufenozide, methoxyfenozide, difluben-

zuron, spinosad, pyriproxyfen, fenpyroximate and etoxazole,

with tebufenozide and methoxyfenozide being the two most

commonly used AIs. Oils included various petroleum oils,

mineral oils and soybean oils. BTs included all registered

strains and the encapsulated delta endotxoin of B. thuringien-

sis. Carbamates included carbaryl, methomyl, aldicarb and

formetanate hydrochloride. Pheromones included (E )-5-

decenyl acetate, (E )-5-decenol, Z-8-dodecenyl acetate,

E,E-8,10-dodecadien-1-ol, E-8-dodecenyl acetate and Z-8-

dodecenol.

Six almond PMPs were identified from the PUR database,

labeled PMP1–PMP6 (Table 1). PMP1 is defined as the tradition-

ally used dormant application of OPs and oil, PMP2 is a combi-

nation of OPs, PYs, and oil, PMP3 combines solely PYs and oil,

PMP4 uses only oils, PMP5 only uses reduced risk controls, BT,

pheromone, or controls listed under ‘‘other’’, and, finally,

PMP6 eliminates dormant season applications all together,

postponing pest treatment until after the rainy season. If

reduced risk pesticides were used in combination with OP

and PYs, the PMP was categorized into PMP1 (if with OP),

PMP3 (if with PYs) or PMP2 (if with both OP and PYs).

Each of these PMPs has strengths and weaknesses in terms

of effectiveness, economics, and environmental impact.

While PMP1 and PMP2 are effective in controlling PTB, SJS,

and aphids, their environmental impact is a problem due to

the potential runoff of OPs. PMP3 eliminates OPs, but is less

effective in controlling SJS. PMP4, which eliminates both OPs

and PYs, is less effective in controlling both PTB and SJS.

PMP5 uses reduced risk products which often are more expen-

sive due to higher material costs and increased numbers of

Table 1 – Almond pest management practices

Pest managementstrategies

Code Description

OP PMP1 Traditional use of OPs in mixture

with oil; effective for PTB, aphids

and SJS.

OPþ PY PMP2 Uses of pyrethroid in combination

with OP and Oil

PY PMP3 Pyrethroid in mixture with oil, no

OP application; not as effective as

OPs for scale control; effective in

most areas for PTB

Oil alone PMP4 Only use oil to control European

red mite eggs, brown almond mite

eggs and low to medium

populations of San Jose scale; this

PMS cannot control peach twig

borer or webspinning mites

Lower risk PMP5 Uses of reduced risk, BT,

pheromone or ‘‘other’’ pesticides

as defined in Table S1 in the

Supplemental Information

No dormant use PMP6 Did not use any insecticides during

dormant season, growers rely on

monitoring and uses of insecticides

during in-season

w a t e r r e s e a r c h 4 2 ( 2 0 0 8 ) 3 6 8 5 – 3 6 9 63688

applications and monitoring. While BT and spinosad are effec-

tive against PTB and moderate levels of SJS, they are less effec-

tive in controlling aphids. Finally, PMP6 eliminates all pest

control during the dormant season, but requires vigilant

monitoring of pest pressure, and may not always be a feasible

option for the grower.

2.5. SWAT model calibration

The Soil and Water Assessment Tool (SWAT) model was used

to simulate the effects of the six PMPs on surface water qual-

ity. This model was developed to predict the impacts of land

management practices on water, sediment and agricultural

chemical yields at watershed scale (Neitsch et al., 2001) and

has been successfully applied to simulate the hydrology of

agricultural watersheds as well as their pesticide and nutrient

loads (Larose et al., 2007; Hu et al., 2007). The SWAT model

uses a number of submodels to predict runoff, pesticide move-

ment, evapotranspiration, sediment transport and nutrient

movement. Seven databases were constructed as inputs:

soil, weather, land use, fertilizer, pesticide, tillage and urban

land use. The output of the model provided information for

predicting the loads of various pesticides on surface water

for each year.

The hydrology and pesticide components of the SWAT

model were accurately calibrated using monitoring data

from the SJR watershed. A detailed description of the calibra-

tion processes was presented in our previous publication (Luo

et al., in press). Daily simulations of the stream flow, sediment

and pesticide loads were calibrated against monitoring data.

The calibrated stream flow, diazinon and chlorpyrifos loads

are shown in Figs. 2 and 3 as examples. The simulation from

the hydrology components matched the actual measure-

ments supplied by the monitoring data quite closely, with

a Nash–Sutcliffe efficiency of 0.96 (Luo et al., in press, Fig. 2).

Although monitoring data for pesticide loads (converted

from concentration) was very limited, the simulated loads

for diazinon and chlorpyrifos appeared to match well with

measurement data from USGS (Fig. 3). The calibrated model

was then used to predict the loads in surface water of the

major OP and PY pesticides: chlorpyrifos, diazinon, permeth-

rin and esfenvalerate, with pesticide use on almonds from

1992 to 2005. Given the non-point source nature of agricultural

water pollution, the simulation was also performed with the

pesticide use data for all the crops within the watershed, in

order to better understand the environmental impacts of

almond PMPs as compared with that of other commodities.

2.6. Data analysis and statistical tests

Based on the timing of applications reported in the PUR data-

base, the PUR data were separated into two seasons: dormant

and in-season. Dormant season was defined as December–

February, while in-season was from March to November.

Data on kilograms of AI and number of treated hectares

were compiled first by each chemical group (OPs, PYs, reduced

risk, BT, pheromone, other) and then grouped into each of the

six PMPs. To understand the influence of the different PMPs on

water quality, a stepwise regression analysis was then used to

test the correlation between PMPs and the chemical concen-

trations predicted by the SWAT model for each season. Chem-

ical concentrations were selected as the dependent variables,

and kg of AI used by each PMP as independent variables. Two

models were chosen to analyze concentrations in water and

sediment, using the four major pesticides (chlorpyrifos, diaz-

inon, permethrin and esfenvalerate), to determine the models

with the best fit. PMP6 was not included in this analysis

because this PMP by definition does not have kg of AI data.

All the data were processed and analyzed using SAS 9.1.2

and Minitab 14.3.

3. Results

3.1. Pesticide use trends

During the dormant season, oil was the pesticide product with

the highest kilograms of active ingredient (1.1� 106 kg per

year), followed by OPs with an average of 2.8� 104 kg per

year. Kilograms of PYs were just below OPs, increasing from

1.0� 103 kg in 2000 to 2.4� 103 kg in 2005. Kilograms of BT

and carbamates decreased dramatically from mid 1990s to

recent years. Reduce risk products showed an increase from

negligible values in 2000 to 2.1� 103 kg of AI in 2005 (Fig. 4).

Analyzing use trends by the number of hectares treated

rather than by total amount offered a slightly different view,

with pyrethroids and OPs changing ranks. While oil again

ranked the highest (4.0� 104 ha per year), PYs exceeded OPs

since 1997 in terms of treated acreage, nearly doubling in

amount from 2000 to 2005 (from 1.0� 104 to 2.1� 104). Hect-

ares treated with OPs ranked third after PY, with little change

over the time period (from 0.4� 104 to 0.7� 104). Similar to the

analysis of total weights, the hectares treated with reduced

risk pesticides showed a gradual increasing trend, ending

0

200

400

600

800

1000

1200

1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

Flo

w (m

3 s

-1)

SimulatedObserved

Fig. 2 – Monthly flow calibration of the SWAT model. The monthly average of stream flow rate was aggregated from daily

data.

w a t e r r e s e a r c h 4 2 ( 2 0 0 8 ) 3 6 8 5 – 3 6 9 6 3689

with 3.8� 104 ha in 2005, which surpassed that of the OP group

in the last year. Very few hectares were treated with BT, carba-

mates, or pheromones (Fig. 5).

In contrast to the chemical group trends, the trends of the al-

mond PMPs experienced more dramatic fluctuations, as seen

over the time period from 1992 to 2005 (Fig. 6). For the dormant

season in 1992, the percent of growers practicing different

PMPs in descending order was PMP1(69.8%)> PMP6

(17.0%)> PMP4(5.3%)> PMP5(4.9%)> PMP3 (2.2%)> PMP2(0.8%);

while in 2005, the order was changed to PMP6(47.3%)>

PMP3(19.3%)> PMP5 (12.9%)> PMP4 (9.9%)> PMP1(9.3%)>

PMP2(1.3%) (Fig. 6). Fig. 6 shows that the use of PMP1 decreased

dramatically by percent of growers, kg of AI, and treated hect-

ares from 1992 to 2002, but increased afterwards. In its place,

PMP6 has become the most popular pest management prac-

tice, in terms of percentage use among growers. In 2005,

47% of almond growers followed PMP6 compared to 17% in

Dia

-50

0

50

100

150

200

250

Lo

ad

(kg

/year)

Y

1992 1993 1994 1995 1996 1997 199

1992 1993 1994 1995 1996 1997 199

0

20

40

60

80

100

Lo

ad

(kg

/year)

Cho

Fig. 3 – Calibration of the SWAT model for simulation of diaz

1992. Use of PMP3 increased by percent of growers, kg of AI

and treated hectares since 2000 and became the dominant

pest management practice in 2005, second only to PMP6.

Users of PMP5 decreased from 5% in 1992 to 1% in 2001 but

increased to 13% in 2005. The uses of PMP5 as measured by

kg of AI and treated hectares also increased in recent years.

The uses of PMP2 and PMP4 did not change much over the

years (Fig. 6).

For in-season pest management practices, more growers

used PMP3 than PMP1 in recent years (Fig. 6). The percentage

of growers using PMP3 during the in-season increased

steadily, from 3% in 1992 to 35% in 2005. A dramatic increase

in the use of PMP5 was observed in both kg of AI and treated

hectares. In 2005, PMP5 was the most-used PMP (excluding

PMP6) by kg of AI. Uses of PMP2 increased greatly in recent

years and became the most-used PMP by treated hectares in

2005 (Fig. 6).

zinon

EAR

PredictionObservation

PredictionObservation

8 1999 2000 2001 2002 2003 2004 2005

8 1999 2000 2001 2002 2003 2004 2005

rpyrifos

inon and chlorpyrifos. Adapted from Luo et al. (in press).

Fig. 4 – Use of pesticides on almonds by amount of AI during dormant season. OP: organophosphate; PY: pyrethroids; Carb:

Carbamates; Reduced Risk: Reduced risk pesticides.

w a t e r r e s e a r c h 4 2 ( 2 0 0 8 ) 3 6 8 5 – 3 6 9 63690

3.2. SWAT simulation outputs

SWAT simulation results clearly showed the temporal trends

of pesticide concentrations in surface water and sediment.

During 1992–2005, water concentrations decreased from

62.2 ppm to 3 ppm for diazinon and from 8.8 ppm to 1.2 ppm

for chlorpyrifos (Fig. 7). Sediment concentrations of these two

chemicals decreased from 15.1� 103 ppm to 0.6� 103 ppm for

diazinon and from 73.7� 103 ppm to 3.0� 103 ppm for chlor-

pyrifos during the same time period (Fig. 7). Both water and

sediment concentrations were lowest in 2003 but increased

in 2002 and 2004 (Fig. 7).

Fig. 7 also shows that permethrin sediment concentrations

peaked at 2.28� 103 ppm in 1998 and 2.56� 103 ppm in 2002,

and then decreased to 0.07� 103 ppm in 2003. Esfenvalerate

05

10152025303540455055

x 103

Treated

A

reas (h

a)

1992 1993 1994 1995 1996 1997 1998

Fig. 5 – Use of pesticides on almonds by treated areas during do

carbamates; Reduced Risk: reduced risk pesticides.

sediment concentrations reached the maximum level of

1.18� 103 ppm in 1998 and decreased to 0.13� 103 ppm in

2003. Water concentrations of these two chemicals showed

a similar pattern having maximum values in 1998 and 2002.

Both water and sediment concentrations increased from

2003 to 2005.

Table 2 shows the percent of use reduction needed to meet

federal and state water quality standards. Current chlorpyrifos

use exceeded EPA and CDFA standards by 2.96% and 11.8%,

respectively. Diazinon use exceeded EPA and CDFA standards

by 1.1% and 2.7%, respectively. To meet the EPA’s standard of

41 ng l�1 and the CDFA’s standard of 15 ng l�1, chlorpyrifos

use needed to be reduced by 88.4% and 95.8% in the SJR water-

shed. Reductions of 61.6% and 78% in diazinon use were

needed to meet the EPA and CDFA standards (Table 2).

Oil OP PYBT Carb Reduced Risk

1999 2000 2001 2002 2003 2004 2005

rmant season. OP: organophosphate; PY: pyrethroids; Carb:

PMP1 PMP2 PMP3PMP4 PMP5 PMP6

020406080

100120140160

0

10

20

30

40

50

60

0

10

20

30

40

50

60

0.0

10.0

20.0

30.0

40.0

50.0

60.0

70.0

80.0

Percen

tag

e o

f g

ro

wers

0102030405060708090

Treated

A

reas (h

a)

0

200

400

600

800

1,000

1,200 x103

x103

x103

x103

Am

ou

nt o

f A

I (kg

)

Dormant In-season

1992

1993

1994

1995

1996

1997

1998

1999

2000

2001

2002

2003

2004

2005

1992

1993

1994

1995

1996

1997

1998

1999

2000

2001

2002

2003

2004

2005

Fig. 6 – Use of different PMPs in almond by percentage of growers, amount of AI, and treated areas during dormant and

in-season.

w a t e r r e s e a r c h 4 2 ( 2 0 0 8 ) 3 6 8 5 – 3 6 9 6 3691

3.3. Impacts of PMP on water quality

The results from the stepwise regression analysis were shown

in Table 3. Only significant correlations were reported. Dor-

mant use of PMP1 was positively correlated with diazinon

and chlorpyrifos concentrations in water and sediment. It

accounted for 72.5% and 78.3% of the variations of diazinon

and chlorpyrifos concentrations in water, and 59.7% and

66.2% in sediment (Table 3). In addition, the use of PMP1 was

negatively correlated with permethrin concentration in water

and sediment contributing to 12.3% and 11.0% of the varia-

tions, respectively. Dormant use of PMP2 was positively asso-

ciated with permethrin and esfenvalerate concentrations in

water and sediment. It accounted for 47.4% and 38.8% of the

variations of permethrin and esfenvalerate concentrations

in water, and 45.9% and 34.7% in sediment. In contrast, dor-

mant uses of PMP3 were negatively correlated with diazinon

concentrations in sediment (5.5%), while PMP4 was negatively

correlated with diazinon in water (4.5%). Finally, PMP5 had

negative correlations with diazinon and chlorpyrifos concen-

tration in both water and sediment, explaining 8.5% and 9.9%

of variations of diazinon and chlorpyrifos in water, and 13.0%

and 6.8% for each chemical in sediment (Table 3).

Most in-season PMPs were not significantly correlated with

any of the independent variables (Table 3). Of those that were

significant, in-season use of PMP1 was negatively correlated

with permethrin in water (13.8%) and sediment (11.5%), while

in-season use of PMP2 had a positive correlation (17.4%) with

esfenvalerate water concentration. All the models were statis-

tically significant with P-values less than 0.05 (Table 3).

0

5

10

15

20

25

Co

ncen

tratio

n in

w

ater (m

g l-1)

X 10-3 Permethrin

Esfenvalerate

0

500

1000

1500

2000

2500

3000

Co

ncen

tratio

n in

sed

im

en

t

(m

g kg

-1)

PermethrinEsfenvalerate

0

10

20

30

40

50

60

70

80 X 103

Co

ncen

tratio

n in

sed

im

en

t

(m

g kg

-1)

DiazinonChlorpyrifos

0

10

20

30

40

50

60

70

Co

ncen

tratio

n in

w

ater

(m

g l-1)

DiazinonChlorpyrifos

1992

1993

1994

1995

1996

1997

1998

1999

2000

2001

2002

2003

2004

2005

1992

1993

1994

1995

1996

1997

1998

1999

2000

2001

2002

2003

2004

2005

1992

1993

1994

1995

1996

1997

1998

1999

2000

2001

2002

2003

2004

2005

1992

1993

1994

1995

1996

1997

1998

1999

2000

2001

2002

2003

2004

2005

Fig. 7 – Simulated concentrations of pesticides in sediment and water.

w a t e r r e s e a r c h 4 2 ( 2 0 0 8 ) 3 6 8 5 – 3 6 9 63692

3.4. The contribution of almond pesticide use to surfacewater pesticide load

Almond pesticide use had a significant contribution to total

pesticide loads in the watershed (Fig. 8). The pesticide loads

at the watershed outlet predicted by the total pesticide use

of all commodities in the regions were dominated by the loads

predicted from pesticide use on almonds. The temporal trends

of pesticide load predicted by almond pesticide use synchro-

nized with that predicted by total commodity use for both

OP and pyrethroid pesticides. In terms of load, majority of

diazinon and chlorpyrifos mass were associated with water

column while majority of permethrin and esfenvalerate

mass were associated with sediment. Therefore, we ran our

correlation test for the OP and PY pesticides in water and

sediment, respectively. Diazinon loads in water contributed

Table 2 – Use reductions to meet water quality standards

No exceeding (%) 5% e

Chlorpyrifos EPA (41 ng l�1)a 88.4

CDFA (15ng l�1)a 95.8

Diazinon EPA (170 ng l�1)a 61.6

CDFA (100 ng l�1)a 78

a The EPA and CDFA criteria are four day average according to CVRWCB (

tration from daily simulation of SWAT model and the chronic water qua

by almond pesticide use were highly correlated with the total

pesticide loads of all commodities, as can be seen by an R2

value of 0.956, while the correlation for chlorpyrifos was not

high with an R2 value of 0.533 (Fig. 8). Predicted permethrin

and esfenvalerate loads in sediment were also highly corre-

lated with almonds and total commodities, with an R2 value

of 0.93 and 0.95, respectively (Fig. 8).

4. Discussion

The presence of pesticides in surface water poses potential

risks to both aquatic species and human (Arias-Estevez

et al., 2008). As an important source of pesticide input to sur-

face water, agricultural pesticide use strategies should be

carefully examined to evaluate their impacts on water quality.

xceeding 10% exceeding Remark (%)

– – Exceeding of current use: 2.96

46.4 1.1 Exceeding of current use: 11.8

– – Exceeding of current use: 1.1

– – Exceeding of current use: 2.7

2006). Percent of exceeding was calculated based on pesticide concen-

lity standards.

Ta

ble

3–

Perc

en

to

fv

ari

ati

on

inch

em

ica

lco

nce

ntr

ati

on

ex

pla

ined

by

pest

ma

na

gem

en

tp

ract

ices

Ch

em

ica

lco

nce

ntr

ati

on

Med

iaD

orm

an

tp

est

ma

na

gem

en

tp

ract

ices

In-s

ea

son

pest

ma

na

gem

en

tp

ract

ices

sta

tist

ics

Sta

tist

ics

PM

P1

PM

P2

PM

P3

PM

P4

PM

P5

PM

P1

PM

P2

PM

P3

PM

P4

PM

P5

R2

(%)

P

OP

OPþ

PY

PY

Oil

alo

ne

Lo

wer

risk

OP

OPþ

PY

PY

Oil

alo

ne

Lo

wer

risk

Dia

zin

on

Wa

ter

72.5

(þþþ

)4.5

(�)

8.5

(�)

85.5

<0.0

01

Sed

imen

t59.7

(þþþ

)5.5

(�)

13.0

(��

)78.3

0.0

01

Ch

lorp

yri

fos

Wa

ter

78.3

(þþþ

)9.9

(�)

88.2

<0.0

01

Sed

imen

t66.2

(þþþ

)6.8

(�)

73.0

0.0

01

Perm

eth

rin

Wa

ter

12.3

(��

)47.4

(þþ

)13.8

(��

)73.6

0.0

03

Sed

imen

t11.0

(��

)45.9

(þþ

)11.5

(��

)68.4

0.0

07

Esf

en

va

lera

teW

ate

r38.9

(þþ

)17.4

(þþ

)56.2

0.0

11

Sed

imen

t34.7

(þþ

)34.7

0.0

08

Perc

en

to

fv

ari

ati

on

(%)¼

seq

uen

tia

lsu

mo

fsq

ua

res

of

on

ev

ari

ab

le/t

ota

lv

ari

an

ce.

(þ):

Po

siti

ve

corr

ela

tio

n.

(�):

Nega

tiv

eco

rrela

tio

n.

Th

en

um

ber

of

‘‘þ

’’a

nd

‘‘�

’’sh

ow

the

stre

ngth

of

corr

ela

tio

n.

(þþþ

):P

erc

en

to

fv

ari

an

ce>

50%

;(þþ

)o

r(��

):p

erc

en

to

fv

ari

an

ceb

etw

een

10%

an

d50%

;(þ

)o

r(�

):p

erc

en

to

fv

ari

an

ce<

10%

.

w a t e r r e s e a r c h 4 2 ( 2 0 0 8 ) 3 6 8 5 – 3 6 9 6 3693

The importance of analyzing pesticide use data and alterna-

tive pest management practices for risk assessment and

effective environmental decision-making was emphasized in

a recent review (Arias-Estevez et al., 2008). This paper pre-

sented the potential of combining the pesticide use surveys

and watershed modeling in the adoption of lower risk PMPs

to replace traditional ones.

The results of this study not only present an informative

snapshot of the environmental impacts on surface water

quality by almond pest management practices but also offer

a basis for recommendations to further reduce impact. The

regression analysis confirmed that PMPs including organo-

phosphates and pyrethroids (PMP1, PMP2) are positively

associated with their concentrations in water and sediment.

While PMP3 did not have any significant association with

pyrethroid concentrations, it is probable that it will in the

future if PMP3 continues to increase in popularity. In con-

trast, the analysis showed that the PMPs using solely oils

(PMP4) or reduced risk products (PMP5) had either negative

association or no association with concentrations of the

four representative chemicals. These results are highly rele-

vant given impending future policy to restrict use of OP and

PY pesticides due to their negative impacts on various non-

target groups.

The analysis of trends on PMP use shows an optimistic

future for reducing impact on water quality. PMP1 has dra-

matically decreased over time in terms of the percentage

of growers employing it, the total kilograms of active in-

gredients, and the total hectares treated, while PMP2

ranked very low by all these measurements. PMP3 is the

greatest concern as its use has been increasing, which

could eventually increase concentrations of pyrethroids

in water systems. However, recent awareness of the

potential adverse effect of pyrethroids on aquatic species

may result in tighter regulations on pyrethroid products,

affecting growers’ preferences for pyrethroid PMPs in the

future. The EPA’s recent decision to re-evaluate over 600

pyrethroids confirmed this surmise. While PMP4 remained

relatively unchanged, PMP5 experienced great increases in

use in recent years. Use of PMP6, which advises no dor-

mant spraying and therefore causes the least negative

environmental impact, has been dramatically increasing

over time. Monitoring pest pressure allowed growers to

take timely action to control pests and to avoid the need

for blanket application of pesticides. The increasing use

of PMP6 indicated that no dormant application was neces-

sary for most of the almond orchards, although during

years with high pest pressure, chemical controls were still

needed. The decreased use of PMP1 together with the sig-

nificant increased use of PMP5 and PMP6 indicates a prom-

ising future for reducing adverse impacts of almond pest

management practices on water quality.

These results are even more promising when taken into

consideration under the larger framework of agricultural

non-point source pollution. The contributions to pesticide

loads from almonds make up a significant proportion of the

total load created by all commodities in the watershed. There-

fore, the positive trends in reducing environmental impact

seen by almond growers will go a long way toward reducing

the overall non-point source pollution in this region. Almond

Chlorpyrifos loadin water

y = 2.1809x + 0.2079R2 = 0.5330

0.0

0.5

1.0

1.5

2.0

2.5

0.0 0.2 0.4 0.6 0.8 1.0

Load by almond pesticide use (mg/s)

Load by almond pesticide use (mg s-1) Load by almond pesticide use (mg s-1)

Load by almond pesticide use (mg s-1)

Diazinon load in water y = 1.3952x + 0.1971R2 = 0.9561

0.001.002.003.004.005.006.007.00

0.00 1.00 2.00 3.00 4.00

Permethrin load in sedimenty = 1.214x + 0.0014

R2 = 0.9299

0.000.010.020.030.040.050.060.070.08

0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07

Esfenvalerate load in sedimenty = 2.5804x + 8E-05

R2 = 0.9495

0.0000.0010.0010.0020.0020.0030.0030.0040.0040.0050.005

0.0000 0.0005 0.0010 0.0015 0.0020Lo

ad

b

y all p

esticid

e u

se

(m

g s

-1)

Lo

ad

b

y all p

esticid

e u

se

(m

g s

-1)

Lo

ad

b

y all p

esticid

e u

se

(m

g s

-1)

Lo

ad

b

y all p

esticid

e u

se

(m

g s

-1)

Fig. 8 – Correlation between pesticide loads predicted by uses of the pesticide on almonds and all crops.

w a t e r r e s e a r c h 4 2 ( 2 0 0 8 ) 3 6 8 5 – 3 6 9 63694

orchards near waterways should be carefully managed with

low risk pest management practices to minimize pesticide

runoff. In orchards with low PWB and SJS pressure, PMP6

may be the best option for reducing impact on water quality.

With careful monitoring and timely application of reduced

risk pesticides during in-season, it is possible to control pests

without any dormant application. It has been demonstrated

that pest damage levels were not higher in the blocks with

no insecticide applications than those in the conventional

blocks which had dormant OP applications (Elliott et al., 2004).

The relationships between pesticide use and loads in sur-

face water are much more complex than an intuitive thinking

that ‘‘higher use resulted in higher pesticide loads’’. Pesticide

loads not only depend on the amount of use, but also the land-

scape characteristics, soil hydrologic related properties, and

many others. For example, factors such as quantity of pesti-

cide use, physical and chemical properties of pesticides, quan-

tity and patterns of rainfall, and timing of pesticide

application affect whether or not and/or how much pesticides

runoff to surface water bodies. Applications of pesticides in

non-vulnerable areas may not pose environmental risks to

water quality. A recent study conducted in vineyards in

a northwest Spain watershed suggested that soils in these

vineyards were not prone to transport fungicides and;

therefore, fungicide use in these vineyards would not pose

any significant risk to water supplies in the study area

(Bermudez-Couso et al., 2007). Timing of pesticide application

and rainfall also play important roles. Uses of pesticides dur-

ing an irrigation season may not have the same impact as the

uses in a rainy season. Our data showed that OP and pyre-

throid concentrations were lowest in 2003 while the total

uses of these pesticides were not. The relative low concentra-

tions were due to a combination of relatively low dormant

pesticide use and low discharge in year 2003.

The percentage of the total pesticide loads that were

contributed by almonds relative to other commodities in

the area differs among pesticides. The differences could be

explained by variation in application timing of these chem-

icals by almonds in comparison to other crops. Fig. 9 shows

that the application timing of chlorpyrifos on almonds and

other crops were synchronized while the application timings

of permethrin were not. Almonds used the majority of per-

methrin during rainy season while other crops used the

majority of permethrin during dry season (Fig. 9). These

different patterns in application timing may result in differ-

ent contributions of pesticide loads associated with the pes-

ticide use on almonds. Pesticides used in rainy season are

generally more prone to runoff, which explains the large

contribution of almonds to total use of permethrin. There

might be other reasons for the differences in almond contri-

bution patterns among chemicals, such as chemical proper-

ties and transport pathways. Hence, further analysis is

needed.

The use of the SWAT model to predict concentrations

of the four representative pesticides in conjunction with

use of pest management practice data presented a novel

method to study the ecological footprint of growers’ farm-

ing practices on water quality. It appears that almond

growers are well on their way to transitioning to lower

impact systems. However, outreach and education to

growers and their pest control advisors can further the

process through assisting in the implementation of low

impact practices and supplementing chemical choices

with best management practices (BMPs) that can mitigate

and/or prevent negative environmental impact. As

growers and policy makers are better informed as to the

ecological footprints of their actions, they will be able to

better transit to more sustainable practices, and the future

0

10

20

30

40

50

60

70

80

Jan Feb March April May June July Aug Sep Oct Nov Dec

x 103

Mo

nth

ly u

se o

f ch

lo

rp

yrifo

s (lb

s o

f A

I)

Chlorpyrifos use on other cropsChlorpyrifos use on almonds

0

1

2

3

4

5

6

7

Jan Feb March April May June July Aug Sep Oct Nov Dec

x 103

Mo

nth

ly u

se o

f p

erm

eth

rin

(lb

s o

f A

I)

Permethrin use on other cropsPermethrin use on almonds

Fig. 9 – Monthly uses of chlorpyrifos and permethrin by almonds and by all commodities in the watershed.

w a t e r r e s e a r c h 4 2 ( 2 0 0 8 ) 3 6 8 5 – 3 6 9 6 3695

of water quality in the San Joaquin River watershed will be

better assured.

5. Conclusion

This paper presented a novel method to study the environ-

mental impacts of different agricultural pest management

practices. By combining the pesticide use surveys and

watershed modeling, we revealed a promising future for

the adoption of lower risk PMPs to replace traditional

ones. While use of traditional dormant OP with oil (PMP1)

decreased dramatically, pyrethroids in combination with

OP and oil (PMP2), pyrethroids in mixture with oil (PMP3),

reduced risk pesticides (PMP5) and no dormant application

(PMP6) have been increasingly used. Although the decrease

of dormant OP use resulted in reduced OP concentrations

in water, the increasing use of pyrethroids in combination

with oil (PMP3) or/and with OP (PMP2) may cause potential

risk to the water quality within the SJR watershed. How-

ever, as reduced risk pesticides and no dormant application

become more and more popular, it is possible to reduce

water quality impact from almond orchards. Since pesti-

cide use on almonds contributes greatly to pesticide loads

in the watershed, especially for permethrin, outreach

efforts are needed to promote better pest management

practices and implement BMPs in almond orchards. This

study suggests an optimistic future of reducing OP pesti-

cide pollution in the SJR by adopting alternative almond

pest management practices.

Acknowledgement

The authors thank the Coalition for Urban/Rural Environ-

mental Stewardship (CURES) and the California State

Water Quality Control Board (CSWQCB) for their financial

support of the research. We also thank Dr. Frank

Zalom for providing materials on his report of ‘‘Alterna-

tives to chlorpyrifos and diazinon dormant sprays’’.

We are very grateful for the reviewer and editor’s

comments.

Supplementary information

Supplementary data associated with this article can be found,

in the online version, at doi:10.1016/j.watres.2008.05.018.

w a t e r r e s e a r c h 4 2 ( 2 0 0 8 ) 3 6 8 5 – 3 6 9 63696

r e f e r e n c e s

Arias-Estevez, M., Lopez-Periago, E., Martınez-Carballo, E., Simal-Gandara, J., Mejuto, J.C.c, Garcıa-Rıo, L., 2008. The mobility anddegradation of pesticides in soils and the pollution ofgroundwater resources. Agriculture, Ecosystems andEnvironment 123 (4), 247–260.

Bacey, J., Spurlock, F., Starner, K., Feng, H., Hsu, J., White, J., Tran, D.M., 2005. Residues and toxicity of esfenvalerate and permethrinin water and sediment, in tributaries of the Sacramento and SanJoaquin Rivers, California, USA. Bulletin of EnvironmentalContamination and Toxicology 74 (5), 864–871.

Bermudez-Couso, A., Arias-Estevez, M., Novoa-Munoz, J.C.,Lopez-Periago, E., Soto-Gonzalez, B., Simal-Gandara, J., 2007.Seasonal distributions of fungicides in soils and sediments ofa small river basin partially devoted to vineyards. WaterResearch 41 (19), 4515–4525.

California Regional Water Quality Control Board Central ValleyRegion, 2006. Amendments to the Water Quality Control Planfor the Sacramento River and San Joaquin River Basins for theControl of Diazinon and Chlorpyrifos Runoff into theSacramento–San Joaquin Delta, June 2006. Final Staff Report.Resolution No. R5-2006-0061, Sacramento, CA, USA.

CDFA and USDA (California Department of Food and Agricultureand U.S. Department of Agriculture), 2006. 2006 CaliforniaAlmond Acreage Report. United States Department ofAgriculture National Agricultural Statistics Service,Sacramento, CA, USA.

CDPR (California Department of Pesticide Regulation), 2006.Summary of Pesticide Use Report Data 2005. CaliforniaDepartment of Pesticide Regulation, Sacramento, CA, USA.

CIMIS (California Irrigation Management Information System),2005. Department of Water Resources. Available from: http://wwwcimis.water.ca.gov/cimis/data.jsp (accessed 03.12.05).

Domagalski, J.L., Dubrovsky, N.M., Kratzer, C.R., 1997. Pesticides inthe San Joaquin River, California-inputs from dormant sprayedorchards. Journal of Environmental Quality 26 (2), 454–465.

Dubrovsky, N.M., Kratzer, C.R., Brown, L.R., Gronberg, J.M.,Buro, K.R., 1998. Water quality in the San Joaquin–TulareBasins, California 1992–1995. In: US Geological Survey Circular,vol. 1159. USGS, Sacramento, CA, USA.

Elliott, B., Wilhoit, L., Brattesani, M., Gorder, N., 2004. PestManagement Assessment for Almonds, Reduced-riskAlternatives to Dormant Organophosphate Insecticides.California Department of Pesticide Regulation, Sacramento,CA, USA.

Epstein, L., Bassein, S., Zalom, F.G., Wilhoit, L.R., 2001. Changes inpest management practice in almond orchards during therainy season in California, USA. Agriculture Ecosystems andEnvironment 83 (1–2), 111–120.

Guo, L., 2003. Semi-empirical Prediction of Pesticide Loading inthe Sacramento and San Joaquin Rivers During Winter StormSeason. Environmental Monitoring, EH03-04. CaliforniaDepartment of Pesticide Regulation, Sacramento, CA, USA.

Hu, X., McIsaac, G.F., David, M.B., Louwers, C.A.L., 2007. Modelingriverine nitrate export from an east-central Illinois watershedusing SWAT. Journal of Environmental Quality 36 (4), 996–1005.

Larose, M., Heathman, G.C., Norton, L.D., Engel, B., 2007.Hydrologic and Atrazine simulation of the Cedar CreekWatershed using the SWAT model. Journal of EnvironmentalQuality 36 (2), 521–531.

Luo, Y., Zhang, X., Liu, X., Ficklin, D., Zhang, M. Dynamicmodeling of organophosphate pesticide load in surface waterin the San Joaquin Valley. Environmental Pollution, in press,doi:10.1016/j.envpol.2008.04.005.

Meister, R., Sine, C., 2005. Crop Protection Handbook. MeisterPublishing Company, Willoughby, Ohio, USA.

Miller, J., 2006. Poisoning Our Imperiled Wildlife, San FranciscoBay Area Endangered Species at Risk from Pesticides. Centerfor Biological Diversity, San Francisco, CA, USA. Report.

NWIS (National Water Information System), USEPA, 2007. Availablefrom: http://waterdata.usgs.gov/nwis (accessed 20.12.07.).

Neitsch, S.L., Arnold, J.G., Kiniry, J.R., Williams, J.R., 2001. Soil andWater Assessment Tool-Version 2000 – User’s Manual.Blackland Research and Extension Center, Temple, Texas, USA.

NRCS (National Resources Conservation Service), USDA 2006.Available from: http://www.ncgc.nrcs.usda.gov/products/datasets/ssurgo/ (accessed 20.05.06.).

Rice, R.E., Jones, R.A., Black, J.H., 1972. Dormant sprays withexperimental insecticides for control of peach twig borer.California Agriculture 26 (1), 14.

Ross, L., Hsu, S.J., White, J., Hefner, K., 1999. Distribution and MassLoading of Insecticides in the San Joaquin River CaliforniaSpring 1991 and 1992. The California Department of PesticideRegulation, Sacramento, CA, USA.

Spurlock, F., 2002. Analysis of diazinon and chlorpyrifossurface water monitoring and acute toxicity bioassay data1991–2001. In: P.M.B (Ed.), Environmental HazardsAssessment. California Department of Pesticide Regulation,Sacramento, CA, USA.

UCIPM (University of California Integrated Pest ManagementProgram), 1985. Integrated Pest Management for Almonds,first ed. Division of Agriculture and Natural Resources,University of California, Oakland, CA, USA. Publication 3308.

USEPA (U.S. Environmental Protection Agency), 2006. ReducedRisk/Organophosphate Alternative Decisions for ConventionalPesticides. US Enviromental Protection Agency Office ofPesticide Programs, Washington D.C. USA. Publication 20460.

USEPA (U.S. Environmental Protection Agency), 2007. Availablefrom: http://www.epa.gov (accessed 12.08.07.).

Werner, I., Deanovic, L.A., Hinton, D.E., 2002. Toxicity ofstormwater runoff after dormant spray application ofdiazinon and esfenvalerate (Asana) in a French prune orchard,Glenn County, California, USA. Bulletin of EnvironmentalContamination and Toxicology 68 (1), 29–36.

Wilhoit, L., Zhang, M., Ross, L., 2001. Data Quality of California’sPesticide Use Report. California Department of PesticideRegulation, Sacramento, CA, USA.

Zalom, F.G., Oliver, M.N., Hinton, D.E., 1999. Alternatives toChlorpyrifos and Diazinon Dormant Sprays. Statewide IPMProject. Water Resoruces Center and Ecotoxicology Program,University of California, Davis, CA, USA.

Zhang, M.H., Wilhoit, L., Geiger, C., 2005. Assessing dormantseason organophosphate use in California almonds.Agriculture, Ecosystems and Environment 105 (1–2), 41–58.